A Puzzle about Hume s Theory of General Representation Abstract According to Hume s theory of general representation, we represent generalities by associating certain ideas with certain words. On one understanding of this theory, calling things by a name does not represent any real qualities of those things or any real relations between them. This interpretation runs into difficulty when applied to Hume s own use of such general terms throughout the Treatise. Because these distinctions do theoreticalexplanatory work in Hume s philosophical system they require that the items so distinguished really are different. This reveals that Hume employs a more sophisticated understanding of the science of human nature than has previously been understood. While Hume is a thoroughgoing nominalist about terms in the language of the vulgar, he is a realist about the theoretical-explanatory terms of science. Keywords: Hume, Representation, Scientific Realism, Impressions, Ideas, Simple, Complex According to Hume s theory of general representation, we represent generalities (such as dog-hood) by associating certain ideas with certain words (like dog ). On one prominent understanding of this theory, calling things by one name or another does not represent any real qualities of those things (their dog-hood) or any real relations between them (that they all resemble each other in some real way that they do not also resemble cats). This interpretation runs into difficulty when we turn our attention to Hume s own use of such general terms throughout the Treatise. It would seem that Hume s own distinctions such as the impressionidea distinction and simple-complex distinction require that the items so distinguished really are different (and are not just associated with different words). This is because these distinctions do an enormous amount of theoretical-explanatory work in the Treatise, work which arguably 1
cannot be done except by assuming that these distinctions represent some real difference between kinds of perceptions. What I will argue here is that this reveals that Hume is working with a more sophisticated understanding of the science of human nature than has previously been understood. In particular, what I will suggest is that while Hume is a thoroughgoing nominalist about terms in the language of the vulgar, he is a kind of scientific realist about the theoreticalexplanatory terms of science (or, as Hume sometimes calls it, the true philosophy). Consider the theory of general representation that Hume presents in section 1.1.7 of his Treatise. According to this theory, roughly, we find certain objects to resemble one another, and so apply a single word to all of them. Upon later hearing this word, we call to mind one or more of these objects. At the same time, we are disposed to recall other ideas that resemble these first ones when appropriately prompted. A particular idea thus made general represents the set of all of the ideas that we are disposed to recall upon hearing a certain word, and the meaning of the word, therefore, also extends to the entire set. This account relies on the notion of our finding certain objects to resemble one another, and this finding can be taken in at least two ways. On the one hand, it might be that we discover the real relation that is capital-r Resemblance between these objects. 1 Call this the Ontological Interpretation. To understand Hume that way is to take him to be, in some sense, committed to an anti-nominalist thesis, namely, the thesis that there is at least one real relation: Resemblance. That is not a particularly attractive way to understand Hume given his announced nominalism, although some have argued that it is unavoidable. 2 Hume explicitly eschews the distinctive ontological commitment that the Ontological Interpretation ascribes to him. On the other hand, a line of interpretation is available according to which this finding of ideas to resemble one another is the not the discovery of a real relation between these items, but is instead 2
nothing over and above our associating those items with each other and with the word (or general term) resemblance. 3 Call this the Nominalist Interpretation. This association is no real relation, but differs only from other associations (e.g. of contiguous items, or of causes and their effects) insofar as we are inclined to associate our resembling ideas with one another, but not also with our contiguous ideas, etc. So, Hume s associations of ideas do not each represent a different real relation that ideas might stand in to one another, but rather only our inclination to parse our ideas into (roughly) three different groups. On this second, and more prima facie attractive, interpretive line, general terms turn out to be, in a sense, arbitrary. While they reflect the tendencies of the human mind to form associations between certain perceptions, this is all that they do. They do not, either in themselves or thereby, reflect any deeper reality. Now consider the place of Hume s theory of general representation within the broader context of the Treatise. That account is essentially situated within his more comprehensive account of the workings of the human mind. Crucial to that account are certain key distinctions that Hume draws both right at the outset of, and throughout, the Treatise: the impression-idea distinction, the simple-complex distinction, the distinction between impressions of sense and impressions of reflection, etc. According to the most thorough version of the Nominalist Interpretation of the theory of general representation, each of these distinctions is, again in some sense, arbitrary: they do not reflect any real relations or qualities of perceptions, but instead only how we happen to associate some, but not others, of these perceptions both with each other and with our names for them. Here, then, is the puzzle. Each of these distinctions is meant to explain some phenomenon of the human mind. The simple-complex distinction, for example, is meant explain the possibility for novel human thought. That explanation, however, has a distinctly, and 3
arguably ineliminable, ontological component: complex perceptions are composed of simple ones. That is, for the simple-complex distinction to do the explanatory work that it does in the Treatise, there must really be a distinction between simple and complex perceptions. The difference between these cannot be merely that some are associated with the term simple and some with the term complex. The latter must be composed of the former. The same will go for Hume s other core distinctions as well. Each is meant to explain some human phenomena; each plays a crucial role within Hume s philosophical system; and each has what will turn out to be an ineliminable ontological component that represents a real difference between the items associated with each term. What I will argue here is that this is enough prima facie evidence for thinking that there is something wrong with the Nominalist Interpretation of Hume s theory of general representation when it is applied to Hume s own theoretical-philosophical distinctions. That is, I will argue that, perhaps despite Hume s own inclinations, we must understand Hume as a kind of scientific realist. While terms in the language of the vulgar must be understood purely nominalistically, the very explanation of why that (and much else) is the case demands that terms in the language of the scientist of man, such as impression or idea, must be understood as robustly ontologically committing, as tracking real qualities of things and real relations between them. This realism, I will suggest, manifests itself in the form of Hume s replacing the language of the vulgar with the language of the scientist of man, and thereby endorsing the latter as a more accurate picture of the way that the world (or at least the world of perceptions) actually is. While I will argue that carefully distinguishing these two modes of speech throughout the Treatise allows us to limit the range of the Ontological Interpretation to its appropriate sphere, even this requires a fairly radical departure from the Nominalist Interpretation. It requires us to 4
understand Hume as working with a much more sophisticated conception of the science of human nature as has previously been attributed to him. That science is not a merely descriptive project, but a theoretical-explanatory one as well, which justifies the employment of ontologically committing theoretical posits. My procedure will be as follows. The devil here is, as it usually is, in the details, specifically in the details of Hume s articulation and treatment of the particular distinctions at issue. So, my method for presenting this puzzle will be to focus on a pair of case studies: first of the simple-complex distinction, and then of the impression-idea distinction. My conclusion in each case will be that doing the explanatory work to which Hume puts them precludes understanding either of these distinctions as the Nominalist Interpretation would have it, but rather seems to require something more like the Ontological Interpretation. As these case studies will take up most of the space of the current paper, I will close by merely outlining the kind of scientific realism that I take the solution of these puzzles to require. Before I turn to those case studies, though, a brief word is in order about the kind of nominalism against which I will argue with respect to certain of Hume s distinctions. 1 Humean Nominalism As I understand the most strict version of Hume s nominalism it is the thesis that the mind is constituted by nothing other than concrete mental particulars perceptions and that the so-called qualities and relations of these are all to be accounted for (or explained away) via appeals to the perceptions themselves and their behaviors. Specifically, according this strictest version of the Nominalist Interpretation, the account of general ideas in 1.1.7 is intended to 5
provide a procedure for doing just this. A perception is blue not because it has the quality blue, but rather because it is associated in the appropriate way with the word blue. Two perceptions are related to one another, say as resembling, not because they stand in the relation of resemblance, but rather because the complex idea of the pair of perceptions is associated in the appropriate way with the word resemblance. This form of nominalism is admittedly extreme, but what I hope to accomplish in this section is to show that this is the form of nominalism to which Hume commits himself (at least with respect to the language of the vulgar), and that scholars attempting to interpret Hume as employing some less strict version of nominalism fail to do justice to Hume s texts and arguments. The most canonical and concise statement of Hume s nominalism comes at T 1.1.7.6 where Hume writes, every thing in nature is individual, but a great deal can be done using other parts of the philosophical system that Hume develops in the Treatise to understand what precisely Hume means by this claim, and to extrapolate his particular use of this thesis to other relevant issues. To begin, though, Hume famously uses this version of his thesis to argue, against Locke, that there can be no such thing as an abstract idea: an idea that is in some way itself indeterminate. While the Individuality Thesis, as we might call it, by itself, does not commit Hume to the very thoroughgoing nominalism attributed to him here i.e. the thesis that the distinctions that we draw as described by Hume s account of general representation do correspond to anything real in objects (or perceptions themselves) what I will argue briefly in this section is that it is plausible to take Hume to be committed to some version of that strong thesis nonetheless. My argument will have three parts. First I will present an argument on Hume s behalf that moves from the Individuality Thesis and premises that Hume more or less explicitly accepts to the denial of all universals and relations. Second I will look at a few places 6
in the Treatise at which Hume seems to deny the strong nominalist thesis, and explain how these are, despite appearances, nonetheless compatible with that thesis. Finally, I will consider the proposal that while Hume may deny the existence of universals and relations, he might remain a nominalist by maintaining that some distinctions correspond to primitive facts about that between which they distinguish. Before I turn to those tasks, however, it is worth pausing to note that Berkeley, whose theory of general representation Hume cites as, one of the greatest and most valuable discoveries that has been made of late years in the republic of letters, (T 1.1.7.1; SBN 17) is explicit in his application of this theory to relations, and Hume may well have simply carried over Berkeley s conclusion in this regard. 4 Hume s stated goal in that section of the Treatise is, to confirm it [Berkeley s theory] by arguments, which I hope will put it beyond all doubt and controversy (T 1.1.7.1; SBN 17). Hume here casts himself as adopting Berkeley s theory and providing extra reasons in support of it. He does not announce a plan either to rehearse that theory in all of its detail or to amend it significantly. So, it would be unsurprising to find Hume discussing only the hard core of the theory, while omitting some of the less central elements e.g. its application to relations. 5 Still, it is easy enough to construct an argument available to Hume from the Individuality Thesis to the conclusion that relations themselves cannot exist, using only premises that Hume more or less explicitly endorses. Here is one way: 1. Everything in nature is individual (particular). 2. Relations must be either particular or universal. 3. If universal, relations do not exist. (1) 7
4. If particular, relations must either exist independently of their relata, or must depend on their relata for their existence 5. For the same reason that the abstract idea of a man cannot exist either with the qualities of all men, or no qualities at all 6 a relation cannot exist without relata. (Such a relation would have to either relate all of its relata, or none of its relata.) 6. Thus, if relations exist, they are inseparable from their relata. (5) 7. The Separability Principle: whatever objects are different are distinguishable and [ ] whatever objects are distinguishable are separable by the thought and imagination (T 1.1.7.3; SBN 18) 8. Thus, if relations exist, they are not different from their relata. (6, 7) 9. Just as the whiteness of a globe is really just the globe itself, relations between objects are just the objects themselves, i.e. only relata, and not relations, exist. (8) 10. Relations do not exist. (3, 9) Thus, combining Hume s stated nominalist thesis the Individuality Thesis with a version of the argument that he gives against abstract ideas, including his use of the Separability Principle, we arrive at precisely the view attributed to him above: that no relations exist. From this, we can derive the thesis of the Nominalist Interpretation as a specification of that principle: relations of resemblance between objects do not exist. So, in addition to the places in the Hume literature cited earlier (n. 2) where scholars have argued persuasively that Hume must hold some such view, we have here a straightforward argument from premises that Hume accepts that very view. Notice that while this argument does not undermine the tenability of the more permissive form of nominalism advocated by Tienson, an additional argument similar to the one above can 8
be given to resist Tienson s proposal. 7 Tienson there distinguishes between skinny universals e.g. some particular shade of a red and fat universals e.g. redness in general and argues that Hume s nominalism consists in the denial of the existence of fat universals combined with a reliance on the existence of skinny ones. Tienson s thought is that since skinny universals are maximally particular, their existence is compatible with the Individuality Thesis. To show that Hume should deny the existence of skinny and fat universals alike, one can use the Separability Principle, as Hume himself arguably does in his discussion of the whiteness and shape of a globe, to show that skinny universals are not distinct from the objects that are supposed to instantiate them, and so are not existences over and above those objects. Much of Tienson s positive argument for understanding Hume as positing the existence of skinny universals depends on his argument that since there must be something that makes true propositions that attribute a property to an object, any form of nominalism that denies the existence of universals is incoherent. Of course, Hume explicitly denies that judgments ever have this form 8 perhaps in part to avoid just such a commitment but even putting that very serious worry aside, notice that this argument simply begs the question against the form of nominalism advocated here. The argument presupposes that the only thing that could make such propositions true is the object s relation to some universal. According to the current line of thought, what makes a proposition attributing a property to a thing true is that that object is a member of the Revival Set of the relevant term. 9 While there is an extended sense in which that is a property of such an object, the hard core of this form of nominalism is that this property is one that is essentially constituted by the associative tendencies of the human mind rather than some worldly item (either the object or its properties). 9
Tienson also discusses two important places in the Treatise where Hume appears to endorse the existence of at least some properties or relations. Given the topic of the current study the relations of resemblance that are the foundation of Hume s theory of general representation perhaps the most important text to address in this regard is the footnote to 1.1.7.6, which occurs in the context of Hume s discussion of general representation and seems to address precisely the issue of the relations of resemblance among, in this case simple, ideas. So, consider that footnote: <ext> Tis evident, that even different simple ideas may have a similarity or resemblance to each other; nor is it necessary, that the point or circumstance of resemblance shou d be distinct or separable from that in which they differ. Blue and green are different simple ideas, but are more resembling than blue or scarlet; tho their perfect simplicity excludes all possibility of separation or distinction. (T 1.1.7.6n; SBN 635) </ext> What Hume appears to claim here is that simple ideas, say of blue and green, can bear real resemblance relations to each other: simple ideas of blue and green resemble each other (with respect to color), while neither so resembles a simple idea of scarlet. It is such relations that will come to form the foundation for the theory of general ideas: we find certain ideas to resemble each other, but not others, call these by the same name, etc. The Nominalist, however, can grant that Hume here claims that simple ideas can resemble each other, and that it is this resemblance that grounds the theory of general 10
representation. His question will then be, How ought one to understand this resemblance? What is at issue between the Nominalist Interpreter and the Ontological Interpreter is not whether Hume claims that certain objects resemble each other, but how to understand this claim. The Ontological Interpreter claims that we must understand it as presupposing real relations of resemblance; the Nominalist Interpreter claims that we must understand it as a shorthand for theideas-that-form-the-revival-set-for-the-term- resemblance. In fact, the Nominalist Interpreter will point to the explanatory power of his interpretation at precisely such moments as the one presented in this footnote. The Ontological Interpreter must understand Hume as appealing to a brute primitive fact about the world: certain items just resemble one another, no more explanation is possible or necessary. 10 That is a very odd sort of thing for Hume of all people to appeal to. The Nominalist Interpreter can claim to do better. What explains the resemblance of blue to green is not a brute fact about blue and green, but rather the psychological fact about us, that we take blue and green to be more similar to each other than we do either of them to scarlet. An explanation that terminates in a fact about human psychology is much more in line with other parts of the Treatise, than is one that terminates in brute ontological facts about the way the world is. Similarly, in the second set of passages that Tienson considers, understanding those passages depends on how one understands the nature of the relations the existence of which Hume appears to concede. Here is Tienson s presentation of this evidence. <ext> In the Section of the Treatise entitled "Of Relations" (I,i,5) Hume distinguishes between "philosophical relations" and "natural relations." Philosophical relations, he tells us, are all of the 11
ways objects can be compared. A natural relation is a "quality, by which two ideas are connected together in the imagination" (Treatise I, i, 5; p. 13). When Hume says philosophical relations are ways objects can be "compared," he does not mean to make them subjective; he means, "admit of comparison." He says, for example, "All kinds of reasoning consist in nothing but a comparison, and a discovery of those relations, either constant or inconstant, which two or more objects bear to each other" (Treatise I,iii,2; p. 73). Comparison discovers relations; it does not produce them. Tienson, 1984: 321 </ext> As Tienson notes, it would be easy enough to understand Hume s talk of comparison here as referring to a subjective activity, if it were not for the passage that he cites at the end here where Hume does write about the discovery of those relations. It is, therefore, worth taking a closer look at this latter passage. <ext> All kinds of reasoning consist in nothing but a comparison, and a discovery of those relations, either constant or inconstant, which two or more objects bear to each other. This comparison we may make, either when both the objects are present to the senses, or when neither of them is present, or when only one. When both the objects are present to the senses along with the relation, we call this perception rather than reasoning; nor is there in this case any exercise of the thought, or any action, properly speaking, but a mere passive admission of the impressions thro the organs of sensation. According to this way of thinking, we ought not to receive as reasoning any of the observations we may make concerning identity, and the relations of time and place; 12
since in none of them the mind can go beyond what is immediately present to the senses, either to discover the real existence or the relations of objects. Tis only causation, which produces such a connexion, as to give us assurance from the existence or action of one object, that twas follow d or preceded by any other existence or action; nor can the other two relations be ever made use of in reasoning, except so far as they either affect or are affected by it. T 1.3.2.2; SBN 73-74 </ext> From what follows the sentence that Tienson quotes, it is clear that the relations that Hume here writes about our discovering are only causal relations. It is reasoning that engages in comparison, and thereby discovers such relations, and the point of this passage is to argue that the reasoning at issue here is probable reasoning, or reasoning from cause to effect and vice versa. Of course, when it comes to causal relations, Hume is notorious for casting these as a mental phenomenon of a kind that fits nicely with Nominalist Interpretation: what one finds when one investigates the idea of causation is not a worldly relation between objects, but rather a custom or habit of the human mind. So, the passage on which Tienson s textual case depends supports his case only if one endorses a highly controversial interpretation of Hume on the reality of causal relations. 11 Like Tienson, Vision also takes Hume to be committed to the reality of at least some relations and qualities, in this case to abstract particulars, which are much like Tienson s skinny universals. <ext> 13
Examples of simple ideas given by Hume include scarlet, orange, sweet and bitter. If a simple idea of sight or touch is a particular idea of, say, a quality separated from other qualities (and not merely attended to separately), then it is an abstract particular of just the type Berkeley execrates. Vision, 1979: 531 </ext> Vision s thought is that the kinds of simple ideas that he cites are particular as opposed to general, but still abstract: they are qualities of objects, not objects themselves. This understanding of simple ideas, however, misses much of what is interesting about Hume s ontology of the mind (about which I will have a great deal to say in the following sections). While Vision is correct to notice that we only arrive at the ideas of these simple ideas through a process of distinguishing, we must be careful to keep this process distinct from the process of abstraction that Locke endorses and Hume rejects. It is not that the simple idea of the sweetness of the apple is an abstract idea of a property as distinct from its object, but rather that simple idea is one of the concrete particular ideas that constitute the concrete particular complex idea of the apple. That this is the case is one of the interesting and distinguishing features of Hume s philosophical system. As Vision himself notes, whatever evidence there is from the texts in which Hume appears to commit himself to the reality of skinny universals, abstract particulars, or relations, this evidence must be balanced against passages in which Hume explicitly denies the existence of precisely such things. <ext> 14
Despite this, in the last paragraphs of the section of the Treatise entitled Of Abstract Ideas Hume denies that we can consider the figure of a piece of a marble without its colour. Moreover, in the Enquiry he rejects the distinction which would permit ideas of a primary quality without concomitant (ideas of) secondary qualities. What he is criticizing in both places are prime instance of abstract particulars of just the kind associated with Berkeley s attack. However, I do not see any recourse but to say that here Hume is simply inconsistent with elements more firmly embedded in his total philosophy. Vision, 1979: 532 33 </ext> Along these lines, the Nominalist Interpreter can also point to texts such as the following. <ext> The word Relation is commonly used in two senses considerably different from each other. Either for that quality, by which two ideas are connected together in the imagination, and the one naturally introduces the other, after the manner above-explained; or for that particular circumstance, in which, even upon the arbitrary union of two ideas in the fancy, we may think proper to compare them. (T 1.1.5.1; SBN 13) </ext> Here Hume certainly seems to be saying that relations are entirely psychological phenomena, of just the kind for which the theory of general representation is meant to account. I.e. when we say of two items that they are related, this amounts to nothing more than the claim that they belong to some shared revival set. Thus, Hume s reference to relata being connected together in the 15
imagination and being the result of the arbitrary union in the fancy. In the case of resembling ideas, this would be the revival set of the term resemblance. 12 If we suppose, contra Vision, however, that Hume did indeed take over Berkeley s theory of general representation, including his denial of the existence of abstract particulars and relations, and read his arguments in 1.1.7 as supporting that theory rather than replacing it, we can avoid having to attribute to Hume inconsistencies of the kind that Vision must. One final way of understanding Hume s nominalism as permitting the existence of some form qualities or relations comes from a recent paper by Hakkarainen that argues that Hume is a trope nominalist who holds that all that exists are Tienson s skinny universals, Vision s abstract particulars, or tropes. 13 This position has the advantage of circumventing the earlier argument against Tienson that moves from the inseparability of skinny universals from their objects to the nonexistence of such skinny universals because the trope nominalist is able to accept this inseparability on the grounds that such objects are nothing but a bundle of tropes. As with Tienson and Vision, though, the primary objection to understanding Hume in this way is textual. Hakkarainen bases much of his case on two passages that support his conclusion only when considered apart from the particular contexts in which they occur. The first is a passage from Of the antient philosophy: <ext> Every quality being a distinct thing from another, may be conceiv d to exist apart, and may exist apart, not only from every other quality, but from that unintelligible chimera of a substance. (T 1.4.3.7; SBN 222) </ext> 16
Considered in isolation this passage certainly does make Hume sound like a trope nominalist. The context in which Hume is writing, however, is one in which he is arguing against a form of Aristotelian substance-accident ontology. So, while Hume s idiom here makes it appear that he is arguing for the independence of qualities, instead he is, as he so often does throughout the Treatise, adopting the objectionable vocabulary of his predecessors (the false philosophy) in order to engage their arguments and theses. Hume s real target in this section is substance, and that he makes his argument against this notion by arguing that it is nothing over and above what the Aristotelian calls accidents or qualities, but which he will ultimately understand as perceptions themselves, does not imply that he himself endorses their view of these. The same explanation applies to the second of Hakkarainen s crucial passages, this one from Of the immateriality of the soul: <ext> If instead of answering these questions, any one shou d evade the difficulty, by saying, that the definition of a substance is something which may exist by itself; and that this definition ought to satisfy us: Shou d this be said, I shou d observe, that this definition agrees to every thing, that can possibly be conceiv d; and never will serve to distinguish substance from accident, or the soul from its perceptions. (T 1.4.5.5; SBN 233) </ext> Here what evidence there is for Hakkarainen s conclusion is even less definitive because Hume now explicitly distinguishes the accidents of the antient philosophy from what he calls 17
perceptions. This distinction is important because while Hume bills the argument that he goes on to make that since the soul is nothing distinct from particular perceptions, it is not itself a real existence as applying equally to accidents and substance, it implies that the relation of accidents to perceptions is itself one that should not be taken for granted. As we have seen, in discussing Tienson and Vision, though, much of Hume s consideration of this relation occurs in 1.1.7, and favors the conclusion that accidents (skinny universals, abstract particulars, or tropes) are merely phantom reflections of the associations that constitute general ideas. There is one final way of understanding Hume s nominalism that I will briefly consider before moving on. What the previous proposals all have in common is the reification of some kind of entity relations, universals, tropes that violates the letter or spirit of Hume s commitment to the existence of only perceptions. It is possible, however, to understand Hume as countenancing the reality corresponding to certain distinctions, such as that between resembling and non-resembling perceptions, without having to accept the existence of any such obscure metaphysical entities. For example, one might suppose that Hume simply takes facts about resemblance to be primitive and unexplainable. Appeals to universals, abstract particulars, and tropes are all ways of explaining these primitive facts about resemblance, this line would go, and that is part of where such attempts go wrong. Facts about resemblance cannot be explained in terms of facts about any more obscure entities, but are also not reducible merely to the associative tendencies of the human mind. 14 As I hope will become clear, when it comes to the distinctions that Hume employs in the course of his pursuit of the science of human nature the simple-complex distinction, the impression-idea distinction, perhaps the distinction between resembling and non-resembling perceptions I am mostly sympathetic to this line. I do, however, count it as moving beyond 18
what is allowed by the Nominalist Interpretation. The motivation for taking this broad view of nominalism stems from the importance of the reductive nature of Hume s account of general representation. That is, if general representations consists of only what Hume describes, then supposing that our use of general terms corresponds to real facts is at best superfluous and at worst (if Hume follows Berkeley as I have suggested) contrary to the intended purpose of that account. Thus, I count this line as a version of the Ontological Interpretation because while it may forego talk of universals and relations per se, it nonetheless commits Hume, in some sense, to the reality of the distinctions at hand. That is a thesis to which I will return in the final section of this paper, wherein I turn to what I will argue is Hume s scientific realism. One final point in favor of the Nominalist Interpretation. The theory of general representation that Hume presents in 1.1.7 accounts for our ability to form general representations without forming abstract ideas, ideas that are anything other than fully determinate particulars. The Nominalist Interpreter will point out that it would be odd for this elimination to be paired with, and in fact depend upon, an acceptance of the universals, relations, or real distinctions that would seem to be required by the Ontological Interpretation. I.e. the Ontological Interpreter would have us eliminate abstract ideas, by relying on real relations of resemblance (or the corresponding real universals). While this is not inconsistent, it does at least seem to be an unhappy marriage. Thus I conclude that the Nominalist Interpretation is not only a live option for understanding Hume s position in the Treatise, but enjoys a number of important benefits over the Ontological Interpretation: it maintains consistency between Hume s declared nominalism and the conclusions that can be drawn using it, it is more in line with the spirit of Hume s typical order of explanation (using psychological facts to explain supposed ontological ones), it can 19
account for the relevant texts, and it avoids having to make a special exception for the treatment of the term resemblance, which on the Ontological Interpretation requires, unlike any other term, an account other than the theory of general representation to explain it. From this I further conclude that the Nominalist Interpretation deserves to be privileged wherever it can be as how we understand the nature of the resemblance relation. If, however, the reader remains unconvinced of the viability of the Nominalist Interpretation, so be it. What I hope to show in what follows is that whatever one makes of this approach vis-à-vis the language of the vulgar, it will not suffice as an account of the general terms that Hume employs in his science of human nature. To that end, I will now turn to the first of two case studies: the simple-complex distinction. 2 The Simple-Complex Distinction We will begin with Hume s distinction between simple and complex ideas. My argument here will be roughly as follows. 1. One s phenomenology is always complex along multiple dimensions. 2. Therefore, the revival set for the general term simple idea will consist of complex ideas. 3. Therefore, the general term simple idea, must represent either (a) the complex ideas that constitute its revival set, or (b) something other than merely the members of its revival set. 4. If the general term simple idea represents the complex ideas that constitute its revival set, then the simple-complex distinction loses its explanatory power (e.g. its role in explaining the novelty of human thought). 20
5. If the general term simple idea represents something other than the member of its revival set, then the Ontological Interpretation of at least that distinction must be correct. The more general conclusion that I will draw by the end of this section is that what Hume offers in the opening sections of the Treatise is a theoretical scientific account of experience. Our experience, the phenomena that a science of human nature must explain, is of only complex ideas. The nature of experience is explained in terms of posited simple perceptions that are governed by general laws and principles such as the Copy Principle, the laws of association, etc. This scientific theory of ideas should be contrasted with a descriptive phenomenology. If we restricted ourselves to experiential items that can be introspectively isolated, we would not appeal to simple ideas. Simple ideas are posits, or theoretical entities, because we have no direct experience of them as simple. It does, however, seem in line with an important strand in Hume s philosophy that simple ideas come to be known through a sophisticated scientific analogy with complex ideas. Hume s commitment to simple ideas is grounded not in direct experience of them, but in their explanatory power. That is a fairly radical conclusion up to which we must work our way slowly. To begin our case study of the simple-complex distinction, then, consider the following. Human phenomenology, of the kind that Hume describes, is always and everywhere incredibly complex. Our visual field at any given time is a complex of spatial points; our diachronic experience is a complex of (themselves complex) temporal parts; our experience as a whole is a complex of perceptions delivered by our various sense modalities. The flipside of this observation about the ubiquitous complexity of human phenomenology is that that phenomenology is never simple. That is, it is never the case that one s phenomenology consists entirely of a single simple perception. 15 One s visual field always consists of multiple perceptions arranged spatially; one s 21
diachronic experience always consists of multiple perceptions succeeding one another; one s experience as a whole is always an experience involving perceptions delivered by more than one sense modality. Of course, it is one of Hume s central theses that this complexity is precisely a complexity composed of absolute simples: minima sensibilia. As we have seen, another important thesis of Hume s is that the meaning of any general term is the set of ideas associated with that term (in the way described in 1.1.7). And, of course, simple idea is itself a general term. <ext> And of this we may be certain, even from the very abstract terms simple idea. They comprehend all simple ideas under them. These resemble each other in their simplicity. (T 1.1.7.7, fn. 5; SBN 635) </ext> Here, then, arises a new puzzle: if our phenomenology is always complex, and the term simple idea derives its meaning from being associated with some set of ideas, then it would seem to follow that the ideas from which the meaning of simple idea derives are themselves complex ideas. That is, according to 1.1.7, the meaning of any general term is the set of ideas associated with that term. In the case of the general term simple idea, since our phenomenology is always complex, this set must consist of complex ideas. And, in fact, this is exactly what we find in the Treatise. Consider, for example, one of Hume s proofs that some such minima sensibilia exist. <ext> 22
Put a spot of ink upon paper, fix your eye upon that spot, and retire to such a distance, that at last you lose sight of it; tis plain, that the moment before it vanish d the image or impression was perfectly indivisible. (T 1.2.1.4; SBN 27) </ext> Consider carefully this procedure, and the image that it is meant to produce. One sees a spot on a piece of paper. That is a spatially complex perception. As one moves away from this piece of paper, the spot grows smaller and smaller until it disappears completely. That experience is both spatially and temporally complex. Presumably, one is hearing, smelling, seeing, etc. various background noise during this experience as well. What Hume s experiment presents, then, by way of demonstrating the existence of simple ideas is a spatially, temporally, and sense-modally complex set of experiences. So, insofar as this kind of procedure is what one is meant to recall upon hearing the term simple idea, one is clearly meant to recall a complex idea. 16 <ext> Tho a particular colour, taste, and smell are qualities all united together in this apple, tis easy to perceive they are not the same, but are at least distinguishable from each other. (T 1.1.1.2; SBN 2) </ext> Here the simple ideas that we are meant to observe are the ones derived from the different sense modalities involved in a complex experience of an apple. All of those simple perceptions occur only as parts of the spatially, temporally, and sense-modally complex experience of biting into an 23
apple. None is experienced all on its own, so to speak. So, in picking out such perceptions, simple idea cannot be straightforwardly associated with just simple ideas. Of course, what one wants to say here is that while it may be certain complex ideas that we are meant to recall upon hearing the term simple idea, it is the simple components of these ideas that we are picking out when we do so. We are meant to focus on the simple ideas to which these complex experiences draw our attention. That may well be, but the question before us concerns what the mechanism is by which we are supposed to perform this operation. Hume offers a theory of general representation, and focus and attention are nowhere to be found in it. Insofar as a general representation as Hume describes it, picks out something over and above the set of ideas that one is disposed to recall upon hearing a certain term, such as the simple idea itself, then one has left the preferred Nominalist Interpretation well behind. It would, of course, be too hasty to draw this conclusion just yet. Here one might also be tempted by the following line. While the revival set for simple idea will be composed of complex ideas, it is the simple components of these ideas (and not the complexes of which they are a part) that stand in the associative (resemblance) relations that are constitutive of this revival set. Thus, while the revival set is composed of complex ideas, it is not the composition of the revival set alone that fixes its content, but rather those of its components that stand in the proper associative (resemblance) relations to one another. Thus, because it is the simple components of the complex ideas that fix the content of the revival set for simple idea, this revival set straightforwardly represents simple ideas as such. 17 Again, though, I believe that the framing problem of this paper once again rears its head, and one must ask: in virtue of what do the resemblance relations among the complex ideas that one actually stands disposed to recall upon hearing simple idea hold between the simple 24
components of these ideas rather than the complex ideas themselves? A revival set consists of a number of associated ideas, where associated means that upon recalling one of these ideas one stands disposed to recall the others. The particular association at work in revival sets is resemblance, and it is tempting to think, as above, that the ideas constituting the revival set all resemble each other in some way or other, e.g. in virtue of their simple components. What is at issue here, however, is exactly how to understand these resemblance relations. According to the distinctions that I have been employing, to suppose that the ideas constituting the revival set of simple idea resemble each other in virtue of their simple components, is to deploy the Ontological Interpretation rather than the Nominalist Interpretation because the former supposes that there is some relation of resemblance over and above the mere associations available via a descriptive phenomenology. Limiting ourselves just to such descriptions, we could only say that we recall, or stand disposed to recall, certain complex ideas upon encountering the words simple idea. The temptation here is to suppose that there is something underlying such associations, something that explains that phenomenology. In fact, this is exactly what I will conclude is necessary by end of this paper. Still, at this point in the dialectic, I am supposing that this is a conclusion that, given the arguments for the Nominalist Interpretation in the previous section, we should expect Hume to hope to avoid. So, our question is this: how can Hume rely on absolutely simple ideas to explain experience when we experience only complex ideas, and even the term simple idea itself appears to pick out a revival set of only complex ideas? The key to answering this question is to see that simple ideas are theoretical posits. Hume s science of man (T Intro. 4; SBN xv) is meant to explain various phenomena. And like the natural sciences, Hume s science of human nature will employ in its explanations theoretical objects that are themselves not phenomena and not directly experienced. They are, in a later 25
parlance, unobservables. Hume posits such entities on the grounds that they explain certain phenomena. In particular, Hume posits simple ideas on the grounds that they explain how it is that experience alone can account for the complex ideas that we in fact have. This strategy can be seen as being introduced in the opening passages of Treatise that we have already begun to discuss. In these opening paragraphs, Hume is aiming at the first, crucial, theoretical proposition, the Copy Principle: that all our simple ideas in their first appearance are deriv d from simple impressions, which are correspondent to them, and which they exactly represent (T 1.1.1.7; SBN 4). He first motivates the Copy Principle by affecting to notice that all of his ideas exactly resemble some impression he has experienced. <ext> The first circumstance, that strikes my eye, is the great resemblance betwixt our impressions and ideas in every other particular, except their degree of force and vivacity. (T 1.1.1.3; SBN 2 3) </ext> Upon further reflection he sees that he has many ideas of which this appears to be untrue, offering as examples the imagined New Jerusalem and inaccurately-remembered Paris. <ext> Upon a more accurate survey [ ] I observe, that many of our complex ideas never had impressions, that correspond to them, and that many of our complex impressions are never exactly copy d in ideas. (T 1.1.1.4; SBN 3) </ext> 26
The resemblance between ideas and impressions struck him as a plausible, naive beginning to an account of the origin of these ideas. In fact, once these apparent counterexamples are dealt with, Hume will go on to use the exact resemblance of simple ideas to simple impressions to do an enormous amount of work in the remainder of the Treatise. So, already treating the Copy Principle as true, he suggests that our mental architecture must be more nuanced than one might suppose pre-theoretically. In particular, he now puts to work the distinction between simple and complex ideas that he introduced a few paragraphs previously. <ext> After the most accurate examination, of which I am capable, I venture to affirm, that the rule here holds without an exception, and that every simple idea has a simple impression, which resembles it; and every simple impression a correspondent idea. (T 1.1.1.5; SBN 3) </ext> In line with the interpretation of Hume as realizing that simple ideas are not experienced as such, this passage should not be read as saying that he confirms the Copy Principle by an accurate examination of simple ideas in isolation along with copied impressions that are similarly experienced on their own. Hume is instead noticing that it will be virtually impossible to find a counterexample to the thesis that the imagination can construct complex ideas from postulated simple ideas that are themselves always copied from correspondingly postulated simple impressions. 27
Of course, because simple ideas constitute complex ideas, there is a sense in which all experience is of simple ideas. This sense in which we do experience simples must, however, be brought into sharper contrast with the sense in which we never experience simple ideas as such. Consider, for example, that a pure sample of mercury in a beaker consists of nothing but mercury atoms. Now, when we look at this mercury or plunge a finger into it and feel it, we certainly do not have visual or tactile experience of individual mercury atoms as such. This is so despite the fact that, by assumption, we are seeing or feeling nothing but the mercury atoms insofar as we see or feel the sample of mercury. We experience the mercury as a silvery cold liquid, but (especially before learning about atomistic theory) not as a complex substance consisting of billions of individual atoms of atomic weight 80. The vial mercury consists of mercury atoms, but in seeing or touching it, we do not experience it as a collection of mercury atoms. Analogously, while Hume s science of human nature teaches us that our phenomenology consists of a large number of simple ideas, we do not experience this phenomenology as a collection of simple ideas. Also analogously, just as we do not typically, or ever at all, experience single individual mercury atoms in isolation, but come to posit these as explanations of certain observed phenomena, we also do not ever experience simple ideas in isolation, but come to posit these as explanations of certain observed phenomena (in this case, certain psychological phenomena). 18 One might be tempted here to draw an important disanalogy between the vial of mercury and Hume s phenomenological explorations on the grounds that while we might be wrong about the composition of the mercury, we cannot similarly be wrong about the composition of complex ideas because the latter are available to us through introspection and in this way present themselves as the simple elements of our complex ideas. If what I have been arguing thus far is 28